Integrated fault detection and isolation and fault-tolerant control of nonlinear process systems.

摘要

Chemical process operations rely extensively on highly automated control systems in order to deal with increasingly stringent requirements of safety, environmental sustainability, and profitability. Automation, however, adds a layer of complexity to a chemical process that may lead to additional faults (e.g., failures in the actuators, sensors or in the controllers) potentially causing a host of safety, environmental and economic problems. Management of abnormal situations, such as automation faults, is a major challenge in the chemical and process industries since abnormal situations account annually for at least 10 billion USD in lost revenue in the US alone. Despite the major industrial importance of the problems of detecting, isolating and handling process/control system faults in a unified and efficient manner, these problems have been traditionally addressed in isolation, thereby significantly limiting the range of practical applicability and performance of the available solutions.;This work develops a general and practical framework for the design of automated fault-tolerant control systems that seamlessly integrate the tasks of fault-detection and isolation and control system reconfiguration for fault handling. Working with general nonlinear dynamic models of chemical processes, we design nonlinear dynamic filters that allow for timely detection and isolation of actuator/control system faults using limited plant measurements. The key idea is to design a fault-detection and isolation scheme for nonlinear process systems that decouples the effect of a fault on all process variables except one. This allows fault detection and isolation for nonlinear chemical processes even with highly coupled variables. The nonlinear dynamic filters are coupled with suitable control system reconfiguration strategies which achieve quick fault recovery and guarantee closed-loop system stability. In addition, fault-tolerant control methods are developed to deal explicitly with the practical issues of limited control actuator capacity, model uncertainty and disturbances, measurement noise and sensor faults. We present applications of the proposed fault-tolerant control system design framework to: (a) a chemical plant consisting of two reactors in series, (b) a high recovery reverse osmosis desalination plant, and (c) a gas-phase polyethylene reactor.